After months of work and a lot of issues about benchmarking comparisons (mainly with XenSource), VMware finally releases first public beta of VMmark benchmarking method, introducing some interesting changes since first announcement at VMworld 2006 last November:
- Reduction of the memory footprint from 7GB per tile to 5GB per tile
- Replacement of the database workload with SysBench running against MySQL
- Distribution of an XML-based open-source benchmarking harness for STAF/STAX
To simplify creation of workload tiles VMware is distributing Linux ones (web server, database server and file server, all based on Novel SUSE Enterprise Linux 10) and a detailed benchmarking guide for creating Windows ones and performing the whole measurement.
Enroll for the beta here.